Gender, Stereotype Threat, and Anxiety ... · PDF fileGender, Stereotype Threat, and Anxiety:...

Gender, Stereotype Threat, and Anxiety:

Psychophysiological and cognitive evidence

Jason W. Osborne

Dept. of Curriculum and Instruction, North Carolina State University, Raleigh,

North Carolina, USA

USA

[email protected]

Gender, Stereotype Threat, and Anxiety: Psychophysiological and cognitive evidence

- 110 - Electronic Journal of Research in Educational Psychology. . ISSN. 1696-2095. No 8, Vol 4 (1) 2006, pp: 109 - 138.

Abstract

Introduction. Claude Steele’s stereotype threat hypothesis proposed that negative group

stereotypes increase individual anxiety levels, hurting performance. However, the role of

anxiety in stereotype threat has not been fully explored. This study examined the hypothesis

that experimental manipulation of stereotype threat would influence real-time measures of

physiological arousal and cognitive efficiency in girls and boys taking mathematics tests.

Method. Participants were students at a large public university in the USA. Girls and boys

were randomly assigned to either high or low stereotype threat conditions, and following an

adaptation period, were presented a challenging mathematics task while physiological meas-

ures were recorded. Cognitive processing time was recorded for each test item.

Results. Results showed significant physiological reactance (skin conductance, skin tempera-

ture, blood pressure) as a function of a stereotype threat manipulation. Results also showed

significant differences in cognitive efficiency as a function of stereotype threat.

Conclusion. These findings are consistent with the argument that stereotype threat manipula-

tions either increase or decrease situationally-specific anxiety. These findings hold significant

implications for high-stakes academic testing and other situations.

Keywords: gender, stereotype threat, academic performance, anxiety,

Jason Osborne

Electronic Journal of Research in Educational Psychology. . ISSN. 1696-2095. No 8, Vol 4 (1) 2006, pp: 109 - 138. - 111 -

Introduction

In the United States, students from disadvantaged minority groups tend to score lower on

important academic tasks than Caucasian or Asian students. Students from disadvantaged

minority groups tend to receive lower grades in school (Demo & Parker, 1987; Simmons,

Brown, Bush, & Blyth, 1978); score lower on standardized tests of intellectual ability (Bach-

man, 1970; Herring, 1989; Reyes & Stanic, 1988; Simmons et al., 1978), and graduate from

college with substantially lower grades than Caucasian students (Nettles, 1988). Ogbu and

others have pointed out similar trends for other disadvantaged groups (e.g., Ogbu, 1978;

Whitworth & Barrientos, 1990).

Decades of research have attributed performance gaps to factors such as socioecomonic

status, academic preparation, and educational opportunities. Yet when background factors are

held constant, subsequent achievement is lower for minority students than Caucasian or Asian

students (Jensen, 1980; Ramist, Lewis, & McCamley-Jenkins, 1994). Further, achievement

gaps are not static, nor do they tend to be present at the beginning of schooling, yet by the

sixth year of school there are substantial gaps between Caucasian students and students of

color in the US (Alexander & Entwhistle, 1988; Valencia, 1991, 1997). Further, data from

the Third International Mathematics and Science Study (e.g., Mullis, Martin, Fierros, Gold-

berg, & Stempler, 2000) shows (a) an increasing performance gap in mathematics and science

(particularly physical/mathematical sciences) as students get older, (b) that this effect is per-

vasive across many countries, but also highly variable across countries, and (c) this pattern

holds even for high-performing students and when these results are investigated at the indi-

vidual item level.

Research has also shown gender disparities in mathematics, sciences, engineering, and

technology. While most girls perform as well as boys in general schoolwork in these areas,

particularly at earlier ages (Hyde, Fennema, & Lamon, 1990), disparities arise when the mate-

rial is more advanced and as girls move into high school and college, the career-choosing

years (Armstrong, 1981; Benbow & Stanley, 1980, 1983; Ethington & Wolfe, 1984; Fennema

& Sherman, 1977, 1978). Women are much less likely to enroll in majors that they perceive

to be highly math-focused (LeFevre, Kulak, & Heymans, 1992), even when they score very

high on standardized math achievement tests. For example, Turner and Bowen (1999) report-

ed that boys scoring “very high” on mathematics standardized tests are over-represented in



engineering, math, and the physical sciences, whereas girls scoring “very high” on the same

tests are over-represented in the biological and non-quantitative social sciences. Further, they

conclude that scores on these tests (indicating, presumably, prior preparation in the field of

mathematics) accounts for less than half the variance in choice of major.

Stereotype threat theory suggests that that girls may withdraw from, or avoid these classes

and majors because math-related anxiety can make them aversive, despite high performance

(e.g., Rounds & Hendel, 1980; Spencer, Steele, & Quinn, 1999; Tobias & Weissbrod, 1980).

There are many possible reasons for the aforementioned phenomena, and have often in-

cluded prior preparation, biological / brain/ genetic differences, and social, or psychological

factors (e.g., Benbow & Stanley, 1980, 1983; Eccles, 1987; Levine & Ornstein, 1983). How-

ever, these explanations cannot account for the effects demonstrated in the stereotype threat

literature.

Stereotype Threat

Claude Steele (1992; 1997) argued that these performance gaps are partly attributable to

negative stereotypes concerning group members’ performance. Because of these stereotypes,

group members tend to experience higher anxiety on tasks in the stigmatized domain than

others not subject to these negative stereotypes. This anxiety is due to the constant fear of

being viewed through the lens of the stereotype, of constantly having to fight against being

stereotyped, and to worry that any personal failure will be a confirmation of the negative

group stereotype.

Increased anxiety not only decreases performance on the task at hand, but also makes the

situation aversive to the student leading students to seek escape from the situation either phys-

ically by absenteeism or withdrawal, or psychologically via disidentification (for further elab-

oration on this argument, see Osborne, 1995, 1997; Steele, 1992, 1997). Thus, Steele’s theory

may help us understand the differential performance of girls and boys on high-stakes mathe-

matics tests as well as the propensity for girls to shun math-intensive classes and majors in

college, despite equal or superior preparation.

Jason Osborne


Empirical support for Stereotype Threat

There is support for many aspects of Steele’s stereotype threat hypothesis. Psychological

theory and research supports the basic assumption that awareness of a negative stereotype

increases situational anxiety and evaluation apprehension as the domain becomes more self-

relevant (Goffman, 1963; Howard & Hammond, 1985; Steele & Aronson, 1995). Further,

increased anxiety or arousal can inhibit performance, particularly when that task at hand is

complicated or not automatized (Geen, 1991; Hunt & Hillery, 1973; Michaels, Blommel,

Brocato, Linkous, & Rowe, 1982; Sarason, 1972; Wigfield & Eccles, 1989) through de-

creased cognitive capacity, reticence to respond, attentional deficits, and distracting or intru-

sive thoughts (Geen, 1991; Sarason, 1972). Clawson, Firment, and Trower (1981) observed

this effect when they reported that secondary-school students who report higher anxiety tend

to score lower on achievement tests in general. Given these findings, two students equal in

ability and preparation could show a significant performance disparity if one were to experi-

ence stereotype threat while the other did not. The achievement gaps might therefore be par-

tially attributable to the effects of increased anxiety.

Stereotype threat and test performance. Since Steele’s hypothesis posits a situationally-

specific cause of underperformance, reducing stereotype threat should close the achievement

gaps (all other things being equal). The results of experimental investigations into this phe-

nomenon have been encouraging (e.g., Steele & Aronson, 1995). Although much of the dis-

cussion and evidence pertains to the gap between African-American and Caucasian students,

studies have manipulated stereotype threat in Latino students (Aronson & Salinas, 1997), and

girls and women in math and science (Shih, Pittinsky, & Ambady, 1999; Spencer et al.,

1999).

Varying the perceived applicability of the stereotype has reduced the gender achievement

gap in experimental situations, a finding that is difficult to explain via biological or social

phenomena. Spencer, Steele, and Quinn (1999) reported that when a stereotype was per-

ceived to be unrelated to a task the gender gap was substantially reduced compared to when

the applicability of the stereotype was not undermined (see also Broadnax, Crocker, & Spen-

cer, 1997). Even highly math-proficient males can experience stereotype threat while taking a

math exam. When presenting students with a math test, Aronson, Lustina, Good, Keough,

Steele, and Brown (1999) told the participants that the purpose of the experiment was to un-



derstand why Caucasian students did so poorly on the (particular) exam compared to Asian

students. As expected, Caucasian males in the stereotype threat condition performed signifi-

cantly worse than Caucasian males in a no-threat condition.

An intriguing study by of highly math-talented Asian-American female undergraduates by

Shih, Pittinsky, and Ambady (1999) demonstrated that Asian-American females' performance

on a math achievement test was enhanced when their Asian identity (and hence the positive

Asian and math stereotype) was made most salient, and undermined when their female identi-

ty (and hence the negative female and math stereotype). Importantly, in this study, groups did

not differ on observed motivation, perceptions of test performance, and were not aware that a

particular target identity was being made more salient.

These and other studies support the following assertions: (a) stereotype threat is

situationally-specific and not a trait of a group, (b) stereotype threat is a phenomenon individ-

uals can experience if they are in a situation where there is a salient negative group stereotype

concerning their performance in that domain and the domain is self-relevant, (c) experiencing

stereotype threat is aversive, as subjects in these conditions show evidence of escape attempts,

(d) acceptance of, or belief in the stereotype is not a necessary condition, and (e) that reducing

stereotype threat improves the performance of members of the stigmatized group to the point

where performance is often not substantially different from that of non-stigmatized groups

once background differences are controlled for. This last point, repeatedly demonstrated in

the studies mentioned above, is the main reason why this theory is the focus of much interest

and attention.

Evidence for anxiety as the explanatory mechanism in stereotype threat. There are several

possible explanations for the observed results. Steele and colleagues (Aronson, Quinn, &

Spencer, 1998; Steele, 1997) argue that anxiety explains (mediates) the observed experi-

mental effects summarized above. However, there are other possible explanations.

It could be that a person holds lower expectations or experiences reduced efficacy when

the target of a negative group stereotype. A study by Spencer, Steele, and Quinn (1999) test-

ed three possible mediators (self-reported state anxiety, evaluation apprehension, and self-

efficacy) in a sample of women and men taking a difficult version of a standardized math test.

Results showed that only self-reported anxiety was found to be a partial mediator of the rela-

Jason Osborne


tionship. Further, the results from Shih et al. (1999) help to rule significant differences in

motivation, perceived performance, liking for the test, assessment of test difficulty, or as-

sessment of personal ability.

While much of the research cited above used highly successful college and university stu-

dents as subjects (making these effects more striking given the restricted range in this popula-

tion as opposed to the general public-school K-12 population in the United States), Osborne

(2001) found that anxiety explained between 38.8% and 41.4% of the racial gap in achieve-

ment test scores in a nationally representative sample of high school seniors. Despite the

promising results, all of these studies have used self-reported anxiety, raising concerns regard-

ing interpretation and causality.

Only Blascovich, Spencer, Quinn, and Steele (2001) have attempted to measure the hy-

pothesized mediator, anxiety, directly. In this study, Blascovich et al. assessed mean arterial

pressure (MAP) reactivity of African-American and Caucasian university students in either

low or high stereotype threat conditions. This study showed that African-Americans under

the high-stereotype threat condition demonstrated significantly greater MAP reactivity than

the other three groups. This study provides direct evidence of physiological reactivity of a

particular type (that could be interpreted as indicative of anxiety) while experiencing stereo-

type threat conditions.

Stereotype Threat and Cognitive Efficiency

Eysenck and Calvo’s (1992) Processing Efficiency Theory suggests that as stress or anxi-

ety increases, cognitive efficiency should suffer. Specifically, they argue that anxiety in-

creases task-irrelevant intrusive thoughts that can disrupt the working memory resources and

the efficiency of the cognitive process. As cognitive efficiency drops, performance should

become worse, or good performance should take longer. This effect should be particularly

pronounced when tasks are challenging or performed under a high cognitive load (see also

Baddeley & Hitch, 1974; Derakshan & Eysenck, 1998; Hopko, Ashcraft, Gute, Ruggerio, &

Lewis, 1998; Klein & Boals, 2001).

Similarly, Hasher and Zacks’ (1988) proposed that increasing anxiety leads to more diffi-

culty regulating attention. According to this perspective, while all individuals experience



task-irrelevant or distracting thoughts, more anxious individuals may give these thoughts

more attention than less-anxious individuals, also either harming performance or increasing

time to perform well. Both perspectives, as well as others (e.g., Geen, 1991; Sarason, 1972),

suggest that anxiety can inhibit performance on academic tasks, particularly when those tasks

are challenging (i.e., not automatized or overlearned). Following this, individuals experienc-

ing stereotype threat should not only show evidence of increased physiological arousal, but

also cognitive sequelae such as increased time to successfully complete tasks or decreased

performance on time-limited tasks.

A recent set of studies by Schmader and Johns (2003, Experiment 1) reported that women

laboring under a stereotype threat condition showed substantially reduced short-term memory

capacity relative to women in a low-stereotype threat condition and men. This research, plus

other research on the cognitive and performance effects of anxiety, suggests that stereotype

threat manipulations, if truly manipulating anxiety, should not only produce physiological

reactance, but also measurable differences in latencies on test items. If memory or other cog-

nitive functioning is at least partially impaired, then individuals laboring under high stereo-

type threat conditions should take substantially longer to successfully respond to test items

than others not laboring under these conditions.

The current study

The literature suggests that Steele’s stereotype threat hypothesis might at least partially

explain the achievement gaps. While authors have shown several different ways to manipu-

late stereotype threat (although there are no studies that actually measure stereotype threat,

validating that these manipulations are actually affecting only stereotype threat), and signifi-

cant effects of these manipulations, the mechanism through which stereotype threat works

remains to be explicated. While studies like that of Osborne (2001) and Blascovich et al.

(2001) suggest the viability of the hypothesis that anxiety is the mediator, neither are ideal.

Studies examining anxiety using self-report measures completed following academic activi-

ties raise obvious issues of causality and interpretation.

The Blascovich et al. (2001) article is mostly focused on explaining the prevalence of hy-

pertension in the African-American population, and it is open to debate as to whether MAP

Jason Osborne


reactivity can be interpreted as an indicator of anxiety or not, given the physiological mecha-

nisms controlling MAP (e.g., Brownly, Hurwitz, & Schneiderman, 2000).

The goal of this study was to explore the link between stereotype threat and physiological

indicators of anxiety or arousal.

Hypotheses

All students should show signs of increased arousal while taking an academic test if they

have any psychological investment in that task. Thus, all hypotheses concerning this study

examine change in some variable (e.g., skin conductance) over time that are more dramatic in

one group than in another group.

Following Steele’s stereotype threat (ST) hypothesis, when girls take a challenging math-

ematics achievement test under “High ST” conditions (i.e., when the stereotype of female

inferiority in mathematics is salient, and girls feel relatively disadvantaged in that domain),

anxiety or stress should increase more dramatically than either girls taking the same test under

“Low ST” conditions (i.e., when the stereotype of female inferiority on this task is specifical-

ly debunked), or boys under either condition. Given this, the first prediction is that girls in

the High ST condition should show significantly worse cognitive efficiency than all other

groups.

Physiological reactance should similarly be more dramatic for girls in the High ST condi-

tion than either girls under Low ST condition or boys. Bradley (2000) gives an excellent

overview of the three physiological indicators we are examining, and their underlying neuro-

logical, chemical, and physiological mechanisms. The following sections draw heavily from

that reference.

Heart rate. According to Bradley (2000), heart rate (HR) is affected by both the sympa-

thetic and parasympathetic nervous systems. Unpleasant visual stimuli tend to produce sig-

nificant initial HR deceleration, while pleasant or erotic imagery tend to produce initial accel-

erations. However, HR can be affected by physical fitness, cardiovascular health, hydration,



posture, respiration, and the need for the body to maintain homeostasis and continue life-

sustaining activities. There also appears to be a difference between the effects of visual and

mental imagery or text-prompted emotion. Text-generated fearful imagery has been shown to

produce HR increases that sustain longer than a few seconds.

Bradley (2000) concludes that one can expect heart rate increases to the extent that emo-

tional mental activity is occurring. In the case of this study, Steele’s theory indicates that

when students are laboring under stereotype threat, there should be increased negative mental

activity consistent with anxiety, threat appraisal, or stress. Thus, in the context of this study,

girls in High ST condition should show greater increases in HR than girls in Low ST or boys

in either condition.

Skin Conductance. Skin conductance (SCL) has been characterized as a pure measure of

sympathetic activity, as most of the electrodermal system is controlled exclusively by the

sympathetic nervous system (Bradley, 2000; Dawson, Schell, & Filion, 2000). Other authors

(e.g., Guyton & Hall, 1996) have argued that palmar sweating, where SCL is usually assessed,

might be parasympathetic in nature because it is controlled by a portion of the hypothalamus

under control of the parasympathetic nervous system. Regardless, it is clear that SCL meas-

ured on the palmar surface of the hands varies dramatically with arousal of either a highly

pleasant (e.g., sexual) or highly unpleasant (e.g., violent) nature. SCL changes have also been

noted as a function of anticipation of pleasant or unpleasant stimuli.

There also appears to be a difference in acclimation to pleasant vs. unpleasant stimuli.

Bradley, Kolchakian, Cuthbert, and Lang (1997) showed that reactions to successive positive/

pleasant stimuli attenuate over time, whereas reactions to negative/unpleasant stimuli tend to

Jason Osborne


retain their magnitude (see also Bradley, 2000). Ultimately, if Steele is correct in his hy-

pothesis, SCL should show more dramatic changes in girls testing under High ST conditions

than any of the other groups.

Surface skin temperature. Blood vessels tend to constrict when an individual is coping

with aversive stimuli, and is most clearly a sympathetic nervous system reaction (Bradley et

al., 1997; Brownly et al., 2000). As blood vessels to the skin constrict flow, surface skin tem-

perature will drop moderately, although body and surface skin temperature are slow to change

(relative to SCL and HR) and are more specifically bounded by the physiological needs of the

body. Life sustaining function simply cannot happen outside a certain narrow temperature

range. Thus, girls in High ST conditions should show greater decreases in surface skin tem-

perature (TEMP) at the extremeties than other groups, but these changes will be relatively

mild and relatively slow compared to other variables.

Method

Participants

Participants were recruited from the psychology pool at a large state university. Par-

ticipants were compensated with course credit. In all, 42 males and 58 females participated

in the study. Participants were randomly assigned to condition through a random number

generator, resulting in 50% of males and 50% of the females in each condition. All experi-

menters were female.

Procedure

Sensor placement. In accordance with recommendations from Dawson et al.(2000),

SCL sensors were attached to the volar surfaces of medial phalanges on first two fingers (in-

dex, middle) of the non-dominant hand (as students used the dominant hand for working out

problems and answering questions). Each sensor cavity was filled sufficiently with biopoten-

tial gel created following instructions from Grey and Smith (1984, p. 553). A sensor for heart

rate was similarly attached to the volar surface of the medial section of the ring finger on the



same hand, and surface skin temperature was attached to the center of the back of the nondo-

minant hand via surgical tape. The experimenter then verified the equipment was receiving a

valid signal from each sensor. Participants were asked not to move their non-dominant hand

during the experiment (as excessive movement can cause erroneous readings in the HR moni-

tor).

Acclimation and baseline measurement. Participants spent ten-minutes in an acclimation

period reading popular magazines (e.g., Popular Science, Car and Driver, National Geo-

graphic, Glamour, Time, etc. Magazines were screened for material of a highly charged na-

ture to prevent baseline measurement error.). After the acclimation period, participants spent

the first part of the experiment performing a procedure where they were presented a series of

simple photographs and paintings and asked to determine the most common or dominant

color in the picture. They responded using the 6-button response pad to get used to using it.

During this time the experimenter recorded baseline physiological data.

At the end of the baseline recording period the experimenter verbally administered the

experimental intervention.

Experimental Manipulation. Stereotype threat manipulations are generally subtle. Previ-

ous research has manipulated: (a) student perceptions as to whether the task assesses aca-

demic or intellectual ability or potential (Katz, Roberts, & Robinson, 1965; Steele & Aronson,

1995), (b) perception as to whether the task is diagnostic of ability (Aronson & Tichy, 1997;

Quinn & Spencer, 1996), (c) performance prior to testing to emphasize student mastery and

improvement in the area (Josephs & Schroeder, 1997), (d) test description to explicitly em-

phasize or de-emphasize that the task does not show group differences in performance

(Broadnax et al., 1997; Spencer et al., 1999) and (e) explicitly describing the test as a measur-

ing of malleable intelligence (Aronson & Fried, 1997; Aronson & Tichy, 1997). In this study

stereotype threat was manipulated using the following manipulation (modeled after Spencer et

al., 1999), which manipulates perception of whether the tests produce gender differences:

High stereotype threat: As you may know, there has been some controversy

about whether there are gender differences in math ability. Previous research has of-

ten noted that girls score lower on math tests than boys. We are trying to understand

why this might be. We are going to give you two short math tests that most people

Jason Osborne


find challenging. When students take these tests, girls consistently do worse than

boys. You will have seven minutes to complete the first test. We want you to do

your best on this test. Take it like you would any other—skip questions you can’t an-

swer, feel free to write anywhere on the test. Remember not to move your [nondomi-

nant] hand while taking the test.

Low stereotype threat: As you may know, there has been some controversy about

whether there are gender differences in math ability. Previous research has often

noted that girls score lower on math tests than boys. However, there are many cases

where girls score as well, or better than boys. We have two short math tests that you

will take. Most people find these tests challenging, although these two tests have

never shown gender differences. We are trying to understand why this might be.

You will have seven minutes to complete the first test. We want you to do your best

on this test. Take it like you would any other—skip questions you can’t answer, feel

free to write anywhere on the test. Remember not to move your [nondominant] hand

while taking the test.

Participants then began the 20-item mathematics test.

Mathematics achievement test

20 challenging mathematics items from college entrance exams constituted the mathemat-

ics achievement test, which was administered via computer and responses were recorded on

the response pad. Participants were allowed to spend as much time as they wanted on each

item, but were encouraged to work as quickly as possible. Thus, all participants completed all

items on this test. Test items were presented in a randomized order to eliminate order effects.

This task was designed to be consistently challenging for all participants; in this study, 48%

of the items were correctly answered, on average, indicating that our goal was met. Males

outscored females with an average of 12.01 to 8.16 items correct (F(1, 87)= 22.74, p < .0001).

There were no condition effects.

Equipment and physiological measures

Heart rate, skin conductance, and surface skin temperature was recorded using a Biolog

3992 from UFI. The Biolog is a small, battery-powered monitoring unit that is double-

shielded from electronic interference with the measurement or recording of the data. The



main unit was attached to sensors with long cords so that it could be placed out of sight of the

participant for minimal intrusiveness.

Heart rate. According to Bradley (2000), heart rate (HR) is affected by both the sympa-

thetic and parasympathetic nervous systems. Unpleasant visual stimuli tend to produce sig-

nificant initial HR deceleration, while pleasant or erotic imagery tend to produce initial accel-

erations. However, HR can be affected by physical fitness, cardiovascular health, hydration,

posture, respiration, and the need for the body to maintain homeostasis and continue life-

sustaining activities. There also appears to be a difference between the effects of visual and

mental imagery or text-prompted emotion. Text-generated fearful imagery has been shown to

produce HR increases that sustain longer than a few seconds.

Bradley (2000) concludes that one can expect heart rate increases to the extent that emo-

tional mental activity is occurring. In the case of this study, Steele’s theory indicates that

when students are laboring under stereotype threat, there should be increased negative mental

activity consistent with anxiety, threat appraisal, or stress. Thus, in the context of this study,

girls in High ST condition should show greater increases in HR than girls in Low ST or boys

in either condition.

Heart rate was measured through a UFI model 1020 Infrared Pulse Plethysmograph

(PPG) transducer that detects heart contractions and ejections by changes in the reflectivity of

the skin of the volar surface of the ring finger. It is sampled with 12-bit resolution at 1000Hz

to detect the QRS peak. The time between two successive peaks is defined as the interbeat

interval, and heart rate calculated as an instantaneous beats per minute (BPM) score from that.

The transducer is sensitive to a detection threshold of 0.25 V.

Skin Conductance. Skin conductance (SCL) has been characterized as a pure measure of

sympathetic activity, as most of the electrodermal system is controlled exclusively by the

sympathetic nervous system (Bradley, 2000; Dawson, Schell, & Filion, 2000). Other authors

(e.g., Guyton & Hall, 1996) have argued that palmar sweating, where SCL is usually assessed,

might be parasympathetic in nature because it is controlled by a portion of the hypothalamus

under control of the parasympathetic nervous system. Regardless, it is clear that SCL meas-

ured on the palmar surface of the hands varies dramatically with arousal of either a highly

Jason Osborne


pleasant (e.g., sexual) or highly unpleasant (e.g., violent) nature. SCL changes have also been

noted as a function of anticipation of pleasant or unpleasant stimuli.

There also appears to be a difference in acclimation to pleasant vs. unpleasant stimuli.

Bradley, Kolchakian, Cuthbert, and Lang (1997) showed that reactions to successive positive/

pleasant stimuli attenuate over time, whereas reactions to negative/unpleasant stimuli tend to

retain their magnitude (see also Bradley, 2000). Ultimately, if Steele is correct in his hypoth-

esis, SCL should show more dramatic changes in girls testing under High ST conditions than

any of the other groups.

Skin conductance was measured via a proprietary UFI voltage excitation SCL signal con-

ditioner that runs a constant 0.5 V across Ag-AgCl electrodes attached to the volar surfaces of

the medial phalanges of the subject’s nondominant hand. This is the signal and setup recom-

mended by Lykken and Venables (1971), with positioning recommended by Dawson, Schell,

& Filion (2000). Biopotential contact medium of the type recommended by Grey and Smith

(1984) was used on both electrodes. SCL was sampled at a rate of 10 Hz, with 12-bit resolu-

tion, and is sensitive to changes of 0.1 uMho across a range of 0.1 to 40.95 uMho.

Surface skin temperature. Blood vessels tend to constrict when an individual is coping

with aversive stimuli, and is most clearly a sympathetic nervous system reaction (Bradley et

al., 1997; Brownly et al., 2000). As blood vessels to the skin constrict flow, surface skin tem-

perature will drop moderately, although body and surface skin temperature are slow to change

(relative to SCL and HR) and are more specifically bounded by the physiological needs of the

body. Life sustaining function simply cannot happen outside a certain narrow temperature

range. Thus, girls in High ST conditions should show greater decreases in surface skin tem-

perature (TEMP) at the extremeties than other groups, but these changes will be relatively

mild and relatively slow compared to other variables.

Surface skin temperature was measured by a UFI model 1070SK solid-state skin tempera-

ture transducer. Using surgical tape, the sensor was attached to the center of the back of the

non-dominant hand. It is sampled at 1 Hz with 12 bit resolution, and is sensitive to changes

of 0.1 degrees Centigrade across a range of 0.1 to 409.6 degrees Kelvin.



Other measures

Perceived performance was assessed at the end of the study via a simple question asking

“How do you think you did on these tests?” Responses were gathered on a scale from 1 (very

poorly) to 6 (very well).

Manipulation checks assessed at the end of the study asked two questions: “In general,

how do you think girls do in math?” and “How do you think girls do on the math tests you

took today?” Both items were assessed on a scale from 1 (much worse than boys) to 5 (much

better than boys). There were no significant differences in the first question (as expected).

For the second question, there as a near-significant trend toward individuals in the High ST

condition reporting that girls tend to do less well (mean=2.77) than in the Low ST condition

(mean=3.01, F(1,67)= 2.95, p < .09).

Data processing

All physiological measures (HR, SCL, TEMP) were measured between one and ten times

per second. Data from these channels were aggregated to 30 second intervals for HR and

SCL by averaging all valid measurementsi within each 30-second interval, and TEMP was

aggregated to 60-second intervals due to the fact that this channel was sampled only once per

second.

Hand movements can produce erroneous readings in HR. Therefore, HR measurements

were screened for values substantially outside the individual’s range. Readings that had the

characteristics of being substantially outside this range (more than double or less than half the

preceding values) for only a very brief (less than 3 measurements) period were assumed to be

the result of hand movement. These infrequent scores were replaced by the 30-second mov-

ing average. Neither of the other two channels suffered this propensity toward artifacts, and

although the data were checked for anomalies, none were detected.

Results

Manipulation checks

Perceived test performance correlated significantly with actual performance (percent correct, r

= .51, p < .001), indicating that participants were generally perceiving their performance rela-

tively realistically. Further, as one might expect, test performance and cognitive efficiency

Jason Osborne


(latencies) were significantly related in a curvilinear relationship indicating that as latencies

increased, performance increased to a point, and then decreased with longer latencies (R= .42,

F(1, 50)=5.35, p < .008, prediction equation y’=-1.17 + 0.0004361*latency -0.000000003941*

latency2 where latency ranged from 17023 to 103735 milliseconds with a median of 42239

milliseconds).

Cognitive efficiency

As all items were presented in a different randomized order for each participant, overall

store and time spent on all test items were averaged to form indices of performance and effi-

ciency ( the average number of seconds spent on each item). Taking longer to correctly an-

swer an item indicates poorer cognitive efficiency, and is also an indicator of anxiety (Ey-

senck & Calvo, 1992).

Figure 1.

Response latencies in milliseconds as a function of sex and stereotype threat.

37000

39000

41000

43000

45000

47000

49000

51000

Hi ST girls High ST boys Low St girls Low ST boys

A univariate Analysis of Variance with condition and sex as independent variables and

test latency (in milliseconds) as the dependent variable yielded only a significant sex by con-

dition interaction (F(1,50)= 3.23, p < .04, 2 = .08). This analysis was followed by an a priori

(planned) contrast comparing girls in the High ST- girls condition to the rest of the partici-

pants. This contrast was significant (F(150)= 6.66, p < .01, 2 = .12). As presented in Figure 1,

girls in the High ST condition had significantly higher latencies than the other groups. Girls

who were in the High ST- condition took over 30% longer to answer each individual test item

than girls in the low ST condition, and approximately 28% longer than either of the boys

groups.



These results are congruent with the results reported by Schmader and Johns (2003), and

support the argument that individuals testing under high stereotype threat conditions experi-

ence significant amounts of cognitive interference.

Physiological measures

Statistical analyses for these data proceeded from an initial mixed between-within re-

peated measures analysis of variance, with scores over time as the within-subjects variable

and sex and condition as between-subjects factors.

Baseline data. Data for the baseline observation period showed no significant differ-

ences as a function of condition in raw SCL (all p > .15), HR (all p > .57 except sex, which

was p < .11) , and TEMP (all p > .51 except sex, which was p < .09).

As Dawson, Schell, and Filion (2000, p. 209) note, it is common to adjust skin con-

ductance scores for the range of the individual’s baseline skin conductance because SCL and

related measures can have large individual differences that is thought to be primarily due to

physiological differences in the anatomy of the skin (e.g., thickness of the corneum; similar

arguments are possible for other measures with strong, anatomically-driven individual differ-

ences, such as heart rate, but not relatively constrained measures such as body temperature),

as well as individual differences in hydration at the time of testing (for the seminal discussion

of these issues, see also Lykken & Venables, 1971). Therefore, raw SCL scores are generally

not of specific interest, but rather, individual variation within that particular individual’s range

that is of interest in psychological research. Ben-Shakhar (1985) clarified this further in rec-

ommending the use of within-subject standardized scores as this relies on a more reliable sta-

tistic, a mean score.

Following this recommendation, then, we computed standardized scores to reflect de-

parture from the averaged baseline measurements for SCL and HR, and both measures were

converted to percent change from baseline to account for the fact that each individual has a

different baseline, and that an individual’s baseline can influence the importance of a particu-

lar magnitude of change. For example, a SCL reading of 5.0 uMho that changes to 10.0

uMHO is a 100% increase, whereas a 5-point increase from 15.0 uMho to 20.0 uMho, is only

a 33% increase, and may reflect a less substantial reaction. The average of the last four

minutes of baseline recording for each channel was defined as baseline for that channel for

Jason Osborne


that individual. Each score after that was converted to a percent increase or decrease from

that score ((score-baseline)*100/baseline). This adjusted score reflects the magnitude of de-

viation from that individual’s unique baseline.

Skin conductance. SCL was subjected to a repeated measures ANOVA, with SCL

over time as the dependent variable and sex and condition as the independent variables.

There was a significant main effect of SCL (F(5, 320)= 28.47, p < .0001, 2 = .31), and a signif-

icant between-subjects interaction between sex and condition (F(1,50)= 5.93, p < .004, 2 =

.05) indicating that girls in the High ST condition had higher average SCL than girls in the

Low ST conditions (means= 55.14, and 13.50, respectively) or boys in either condition

(means=13.57, and 22.93, respectively). As expected, there was also a significant SCL by sex

by condition interaction (F(5, 250)= 2.91, p < .002, 2 = .08), depicted in Figure 2. No other

effect was significant.

0

10

20

30

40

50

60

70

80

Bl t1 t2 t3 t4 t5

Hi St F

Hi St M

Lo ST- F

Lo ST- M

Figure 2

Percent Change from Baseline in Skin Conductance (SCL)

This interaction was explored via univariate a priori contrasts comparing girls in the

High ST condition with all other groups. All contrasts were significant at levels ranging from

p < .0001 to p < .02.

Heart rate. The analyses of heart rate yielded no significant effects. Trends were in

the predicted direction, however.

Surface skin temperature. A repeated measures ANOVA indicated a significant effect

for surface skin temperature. Specifically, there was a significant main effect for change in

TEMP over time (F(7, 441)= 29.58, p < .0001, 2 = .32), and a significant interaction of condi-



tion and sex (F(1,50)= 4.17, p < .02, 2 = .12). The interaction of interest was the TEMP by

condition by sex interaction (F(7,350)= 4.33, p < .0001, 2 = .12) indicating that girls in the

High ST condition showed the expected decrease in surface skin temperature, while other

groups showed increases (presented in Figure 3). Post -hoc comparisons revealed significant

contrasts between girls in the High ST condition and other groups for the last 3 time periods

(marked on the Figure with an asterisk). This is understandable as skin temperature is a rela-

tively slow-changing variable compared to heart rate and SCL.

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

BL t1 t2 t3 t4 t5* t6* t7*

Hi ST F

Hi ST M

Lo ST F

Lo ST M

Figure 3

Percent Change from Baseline in Surface Skin

Temperature (60 second intervals)

Discussion

This study was designed to explore Steele’s hypothesis that students subjected to stereo-

type threat might experience relatively inflated levels of anxiety, stress, or arousal, and that

this increased physiological arousal should impair cognitive efficiency. Consistent with ex-

pectations, the results of this study show evidence of physiological reactance for girls under

high stereotype threat conditions that are consistent with an anxiety or autonomic arousal re-

action. Substantial and significant reactance in skin conductance (SCL), surface skin temper-

ature (TEMP), and diastolic blood pressure (DBP) all support Steele’s argument regarding the

nature of stereotype threat, and its possible affects on academic performance. These results

were also congruent with the findings by Blascovich et al. (2001).

Jason Osborne


Further, these results show a plausible link between the manipulation of stereotype threat

and actual academic performance. As multiple theories (discussed above) point out, increased

anxiety should decrease cognitive efficiency, which should thus either cause students to take

longer to perform at par with students not laboring under increased anxiety, or, when tasks are

time-limited, to perform more poorly, despite being equally capable and prepared. This sec-

ond finding holds significant implications for academic testing, as many tests (particularly

high-stakes tests) tend to be time- limited. It is not surprising that in the US there are signifi-

cant performance gaps on high-stakes achievement tests when our results show it takes stig-

matized groups up to 30% longer per item to answer.

One of the important lessons from this and other stereotype threat studies is that under low

stereotype threat conditions, achievement gaps tend to attenuate. Given that these manipula-

tions are often quite subtle (i.e., merely telling students that tasks are fair or specifically de-

bunking stereotypes of inferiority), there is reason to hope that subtle changes to testing para-

digms can substantially improve the opportunity for students from stigmatized groups to per-

form to their potential.

Caveats and directions for future research

It is important to remember that while it is intuitive that anxiety and stress are related to

physiological reactions, there is controversy and debate in the literature regarding the rela-

tionship between physiological changes and changes in emotionality. The safest interpreta-

tion of these results is that physiological reactance was increased under high stereotype threat

conditions, relative to low stereotype threat conditions. Labeling these reactions as indicative

of anxiety is not technically supported by the science of psychophysiology at this point. Nev-

ertheless, these results are consistent with Steele’s stereotype threat hypothesis.

Second, these studies investigated one particular aspect of stereotype threat (sex and math

performance). Although it seems intuitive to generalize to other instances (e.g., race, age),

that should be done with care.

Furthermore, it is reasonable to wonder whether experimenter sex may have had an influ-

ence in the reactions of the participants (following research by Inzlicht & Ben-Zeev, 2000 and

others). In this study all experimenters were female, and thus we held sex constant across



condition. However, this in itself represents an important line of research, especially as stere-

otype threat relate to education. Is it the case, for example, that a female math teacher can

reduce stereotype threat more effectively than a male teacher?

One must note that the participants in this study were University students, and therefore

represent a select segment of the student population most likely to be successful in academics.

However, as many of these participants were psychology majors, many of whom are notori-

ously math-averse, it may not be as difficult a generalization as one might expect. Addition-

ally, this sample, if it represented only the most successful segment of the student population,

would therefore bias the results toward the null hypothesis, rather than advantaging the results

toward rejection of the null. Having found some substantial effects within this restricted pop-

ulation, it makes another study using secondary school students an interesting prospect and

important direction for future research.

These studies did not replicate earlier findings showing that manipulation of stereotype

threat can produce differences in performance on academic tasks. The difference was that

participants had unlimited time on each item. It is likely (given the data on latencies showing

girls in the high -stereotype threat condition took substantially longer to answer than partici-

pants in other conditions) that had this test been time-limited, results would have replicated

previous studies.

Finally, as with much of this type of research, within-group variance was substantial, hurt-

ing statistical inference. Some individuals in high stereotype threat conditions showed virtu-

ally no reaction, while others responded dramatically. Likewise, some individuals in the low

stereotype threat condition showed reactance where others did not. This field and this para-

digm needs to move to a more individualistic approach to understanding the effects of stereo-

types on an individual level. Future research needs to look at what individual differences pre-

dict whether an individual will react or be sensitive to stereotypes, and what individual differ-

ences might be protective against these effects. Researchers have identified several variables

that are related to individual differences in reactance to stereotype threat. These include

stereotype relevance (Brown & Josephs, 1999), gender identification (Schmader, 2002), and

others, including identification with academics (e.g., Steele, 1997). Identification with aca-

demics was assessed in both studies, but unfortunately the college populations participating in

the study had a relatively restricted range in identification with academics (all scoring rela-

Jason Osborne


tively high). Thus, moderation effects could not be examined, and covarying identification

with academics failed to produce any alteration in the nature of the results. A more sensitive

measure of identification with mathematics performance would have been desirable (as used

in Brown & Josephs, 1999), and future research should attempt to measure domain identifica-

tion as sensitively and narrowly as possible.

However, not having this measure, and thus, not being able to separate out less identified

from more identified individuals will only add error variance to the data, decreasing the like-

lihood of rejecting the null hypothesis. Since significant trends were observed, although un-

fortunate, this is not a fatal flaw.

Conclusions

In the USA it is virtually compulsory for states to participate in high-stakes testing of stu-

dents from very early in their academic career. Given the stereotype threat literature, and

other theoretical and methodological reasons that contraindicate the use of high-stakes testing

in public education, educators are left in an interesting dilemma. If stereotype threat research

is to be taken seriously, then one must seriously consider the notion that widely-reported

“achievement gaps” between various groups may be little more than the effect of relatively

empoverished backgrounds and societal stereotypes. This requires us to carefully question

high-stakes testing in education.

On the positive side, stereotype threat research has demonstrated that minimal alterations

to testing situations can substantially reduce the observed achievement gaps, at least in re-

search settings. This should be pursued in order to engineer testing and classroom situations

to minimize stereotype threat, or maximize student resilience in the face of ubiquitious nega-

tive group stereotypes. It has now been almost 15 years since Steele’s (1992) Atlantic

Monthly article announced his formulation of his Stereotype Threat hypothesis. The literature

has demonstrated the importance of the idea. Implementation of the principles and lessons

from a decade of research cannot come quickly enough.



References

Alexander, K. L., & Entwhistle, D. R. (1988). Achievement in the first two years of school:

Patterns and processes. Monographs of the Society for Research in Child Develop-

ment, 53(2).

Armstrong, J. M. (1981). Achievement and participation of women in mathematics: Results

from two national surveys. Journal of Research in Mathematics Education, 12, 356-

372.

Aronson, J., & Fried, C. (1997). Belief in the malleability of intellectual ability: An interven-

tion to increase the performance of African-American college students.Unpublished

manuscript.

Aronson, J., Lustina, M. J., Good, C., Keough, K., Steele, C. M., & Brown, J. (1999). When

White men can't do math: Necessary and sufficient factors in stereotype threat. Jour-

nal of Experimental Social Psychology, 35(1), 29-46.

Aronson, J., Quinn, D. M., & Spencer, S. J. (1998). Stereotype threat and the academic un-

derperformance of minorities and women. In J. K. Swim & C. Stangor (Eds.), Preju-

dice: The Target's Perspective. New York: Academic Press.

Aronson, J., & Salinas, M. F. (1997). Stereotype threat, attributional ambiguity, and Latino

underperformance.Unpublished manuscript.

Aronson, J., & Tichy, J. C. (1997). Conceptions of ability and stereotype threat.Unpublished

manuscript.

Bachman, J. G. (1970). Youth in Transition: Volume 2. The impact of family bachground

and intelligence on 10th-grade boys. Ann Arbor, Michigan: Survey Research Center,

Institute for Social Research.

Baddeley, A. D., & Hitch, G. (1974). Working Memory. In G. A. Bower (Ed.), The psychol-

ogy of learning and motivation (Vol. 8, pp. 47-89). New York: Academic Press.

Benbow, C. P., & Stanley, J. C. (1980). Sex differences in mathematical ability: Fact or arti-

fact? Science, 210, 1262-1264.

Benbow, C. P., & Stanley, J. C. (1983). Sex differences in mathematical reasoning ability:

More facts. Science, 222, 1029-1031.

Blascovich, J., Spencer, S. J., Quinn, D., & Steele, C. (2001). African Americans and high

blood pressure: The role of stereotype threat. Psychological Science, 12(3), 225-229.

Bradley, M. M. (2000). Emotion and Motivation. In J. T. Cacioppo, L. G. Tassinary & G. G.

Berntson (Eds.), Handbook of Psychophysiology, 2nd Edition. Cambridge, UK: Cam-

bridge University Press.

Jason Osborne


Bradley, M. M., Kolchakian, M., Cuthbert, B. N., & Lang, P. J. (1997). What's new? What's

exciting? Novelty and emotion in perception. Psychophysiology, 34, 1-22.

Broadnax, S., Crocker, J., & Spencer, S. J. (1997). African-Americans and academic per-

formance: The medicational role of stereotype vulnerability. Unpublished manuscript.

Brown, R. P., & Josephs, R. A. (1999). A burden of proof: stereotype relevance and gender

differences in math performance. Journal of Personality and Social Psychology,

76(2), 246-257.

Brownly, K. A., Hurwitz, B. E., & Schneiderman, N. (2000). Cardiovascular psychophysiol-

ogy. In J. T. Cacioppo, L. G. Tassinary & G. G. Berntson (Eds.), Handbook of Psyho-

physiology, 2nd Edition. Cambridge, UK: Cambridge University Press.

Clawson, T. W., Firment, C. K., & Trower, T. L. (1981). Test anxiety: Another origin for ra-

cial bias in standardized testing. Measurement and Evaluation in Guidance, 13, 210-

215.

Dawson, M. E., Schell, A. M., & Filion, D. L. (2000). The electrodermal system. In J. T. Ca-

cioppo, L. G. Tassinary & G. G. Berntson (Eds.), Handbook of Psychophysiology, 2nd

Edition. Cambridge, UK: Cambridge University Press.

Demo, D. H., & Parker, K. D. (1987). Academic achievement and self-esteem among Afri-

can-American and White college students. Journal of Social Psychology, 127(4), 345-

355.

Derakshan, N., & Eysenck, M. W. (1998). Working memory capacity in high trait-anxious

and repressor groups. Cognition and emotion, 12(5), 697-713.

Eccles, J. S. (1987). Gender roles and women’s achievement-related decisions. Psychology of

Women Quarterly, 11, 135-172.

Ethington, C. A., & Wolfe, L. M. (1984). Sex differences in a causal model of mathematics

achievement. Journal for Research in Mathematics Education, 15, 361-377.

Eysenck, M. W., & Calvo, M. G. (1992). Anxiety and performance: The processing effi-

ciency theory. Cognition and emotion, 6, 409-434.

Fennema, E. H., & Sherman, J. A. (1977). Sex-related differences in mathematics achieve-

ment, spatial visualization, and sociocultural factors. American Educational Research

Journal, 14(51-71).

Fennema, E. H., & Sherman, J. A. (1978). Sex-related differences in mathematics achieve-

ment: A further study. Journal for Research in Mathematics Education, 9, 189-203.

Geen, R. G. (1991). Social Motivation. Annual Review of Psychology, 42, 377-399.



Goffman, E. (1963). Stigma: Notes on the management of a spoiled identity. New York:

Touchstone.

Grey, S. J., & Smith, B. L. (1984). A comparison between commercially available electrode

gels and purpose-made gel in the measurement of electrodermal activity. Psycho-

physiology, 21, 551-557.

Guyton, A. C., & Hall, J. E. (1996). Textbook of Medical Physiology. Philadelphia, PA.:

Saunders.

Hasher, L., & Zacks, R. T. (1988). Working memory, comprehension, and aging: A review

and a new view. The Psychology of Learning and Motivation, 22(193-225).

Herring, R. D. (1989). Minimum competency testing: Implications for ethnicity. Journal of

Multicultural Counseling and Development, 17, 98-104.

Hopko, D. R., Ashcraft, M. H., Gute, J., Ruggerio, K. J., & Lewis, C. (1998). Mathematics

anxiety and working memory: Support for the existence of a deficient inhibition

mechanism. Journal of Anxiety Disorders, 12(4), 343-355.

Howard, J., & Hammond, R. (1985, September 9). Rumors of inferiority. The New Republic,

193, 17-21.

Hunt, P. J., & Hillery, J. M. (1973). Social facilitation in a location setting: An examination

of the effects over learning trials. Journal of Experimental Social Psychology, 30, 139-

155.

Hyde, J. S., Fennema, E., & Lamon, S. J. (1990). Gender differences in mathematics perform-

ance: A meta-analysis. Psychological Bulletin, 107, 139-155.

Inzlicht, M., & Ben-Zeev, T. (2000). A threatening intellectual environment: Why females are

susceptible to experiencing problem-solving deficits in the presence of males. Psycho-

logical Science, 11, 365-371.

Jensen, A. R. (1980). Bias in Mental Testing. New York: Free Press.

Josephs, R., & Schroeder, D. (1997). The self-protective function of the learning

curve.Unpublished manuscript.

Katz, I., Roberts, S. O., & Robinson, J. M. (1965). Effects of task difficulty, race of adminis-

trator, and instructions on digit-symbol performance of Negroes. Journal of Personal-

ity and Social Psychology, 2, 53-59.

Klein, K., & Boals, A. (2001). The relationship of life events stress and working memory ca-

pacity. Applied Cognitive Psychology, 15, 565-579.

Jason Osborne


LeFevre, J., Kulak, A., & Heymans, S. (1992). Factors influencing the selection of university

majors varying in mathematical content. Canadian Journal of Behavioural Science,

24, 276-289.

Levine, D. U., & Ornstein, A. C. (1983). Sex differenecs in ability and achievement. Journal

of Research and Development in Education, 16, 62-72.

Lykken, D. T., & Venables, P. H. (1971). Direct measurement of skin conductance: A pro-

posal for standardization. Psychophysiology, 8, 656-672.

Michaels, J. W., Blommel, J. M., Brocato, R. M., Linkous, R. A., & Rowe, J. S. (1982). So-

cial facilitation and inhibition in a natural setting. Replications in Social Psychology,

2(21-24).

Mullis, I.V., Martin, M. O., Fierros, E. G., Goldberg, A. L., & Stemler, S. E. (2000). Gender

differences in achievement: IEA’s Third Internacional Mathematics and Science

Study (TIMMS). TIMMS International Study Center, Boston College, Chestnut Hill:

MA.

Nettles, M. T. (1988). Toward undergraduate student equality in American higher education.

New York: Greenwood.

Ogbu, J. U. (1978). Minority education and caste: The American system in cross-cultural

perspective. New York: Academic Press.

Osborne, J. W. (1995). Academics, self-esteem, and race: A look at the assumptions underly-

ing the Disidentification hypothesis. Personality and Social Psychology Bulletin,

21(5), 449-455.

Osborne, J. W. (1997). Race and academic disidentification. Journal of Educational Psychol-

ogy, 89(4), 728-735.

Osborne, J. W. (2001). Testing stereotype threat: Does anxiety explain race and sex differ-

ences in achievement? Contemporary Educational Psychology, 26(3), 291-310.

Osborne, J. W. (2004). Identification With Academics and Violence in Schools. Review of

General Psychology, 8(3), 147-162.

Quinn, D., & Spencer, S. J. (1996). Stereotype threat and the effect of test diagnosticity on

women's math performance. Paper presented at the American Psychological Associa-

tion, Toronto, Canada.

Ramist, L., Lewis, C., & McCamley-Jenkins, L. (1994). Student group differences in predict-

ing college grades: Sex language and ethnic groups (College Board Report No. 93-1,

ETS No. 94.27). New York: College Entrance Examination Board.



Reyes, L. H., & Stanic, G. M. (1988). Race, sex, socioeconomic status, and mathematics.

Journal for Research in Mathematics Education, 19, 26-43.

Rounds, J. B., & Hendel, D. D. (1980). Measurement and dimensionality of mathematics

anxiety. Journal of Counseling Psychology, 37, 138-149.

Sarason, I. G. (1972). Experimental approaches to test anxiety: Attention and the uses of in-

formation. In C. D. Spielberger (Ed.), Anxiety: Current trends in theory and research

(Vol. 2). New York, NY: Academic Press.

Schmader, T. (2002). Gender identification moderates stereotype threat effects on women's

math performance. Journal of Experimental Social Psychology, 38, 194-201.

Schmader, T., & Johns, M. (2003). Converging evidence that stereotype threat reduces work-

ing memory capacity. Journal of Personality and Social Psychology, 85, 440-452.

Shih, M., Pittinsky, T. L., & Ambady, N. (1999). Stereotype susceptibility: Identity salience

and shifts in quantitative performance. Psychological Science, 10, 80-83.

Simmons, R. G., Brown, L., Bush, D. M., & Blyth, D. A. (1978). Self-esteem and achieve-

ment of Black and White adolescents. Social Problems, 26, 86-96.

Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat and women's math

performance. Journal of Experimental Social Psychology, 35(1), 4-28.

Steele, C. M. (1992, April). Race and the schooling of Black Americans. The Atlantic

Monthly, 68-78.

Steele, C. M. (1997). A threat in the air: How stereotypes shape intellectual identity and per-

formance. American Psychologist, 52(6), 613-629.

Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual test performance of

African Americans. Journal of Personality and Social Psychology, 69(5), 797-811.

Tobias, S., & Weissbrod, C. (1980). Anxiety and mathematics: An update. Harvard Educa-

tional Review, 50(1), 63-80.

Turner, S. E., & Bowen, W. G. (1999). Choice of Major: The Changing (Unchanging) Gen-

der Gap Source. Industrial & labor relations revie, 52(2), 289-313.

Valencia, R. R. (1991). The plight of Chicano students: An overview of schooling conditions

and outcomes. In R. R. Valencia (Ed.), Chicano school failure and success: Research

and policy agendas for the 1990s (pp. 3-26). London, England: Falmer Press.

Valencia, R. R. (1997). Latinos and education: An overview of sociodemographic character-

istics and schooling conditions and outcomes. In M. Barrera-Yepes (Ed.), Latino edu-

cation issues: Conference proceedings. Princeton, NJ: Educational Testing Service.

Jason Osborne


Whitworth, R. H., & Barrientos, G. A. (1990). Comparison of Hispanic and Anglo graduate

record examination scores and academic performance. Journal of Psychoeducational

Assessment, 8(2), 128-132.

Wigfield, A., & Eccles, J. S. (1989). Test anxiety in elementary and secondary school stu-

dents. Educational Psychologist, 24, 159-183.

----------------------- 1 In the case of some physiological measurement, there are occasionally missed or dropped

measurements due to things such as participant finger movement. These are fairly rare, and in

this study were primarily missed heart rate measurements.



[This page intentionally left blank]

Gender, Stereotype Threat, and Anxiety ... · PDF fileGender, Stereotype Threat, and Anxiety:...

Documents

Transcript of Gender, Stereotype Threat, and Anxiety ... · PDF fileGender, Stereotype Threat, and Anxiety:...